Here's the magic script.  Basically what it does is create a seperate
schema for the national shapefiles and then each state.  Then in the
state schemas there are tables which are state-level shapefiles, and
then county-level shapefiles which are indicated with a 'c999_' prefix.

It uses shp2pgsql from PostGIS and dbview (from their respective Debian
packages).  There are a bunch of variables at the top that you'll need
to customize, of course.  Also, there are a set of parameters it takes
which basically tell it what you want to load.

-n     Load national-level data
-b     Load state-level data
-c     Load county-level data

-s <2-digit state code>  Individual state code to load
-c <3-digit or 5-digit county code>  Individual county code to load

-d     Drop tables/etc before creating/loading the data

The script also needs a set of 'helper' SQL files, which I've included
in the 'tiger_sql.tar.gz'.  These should be put under an 'sql' directory
under wherever you're loading the data from.

That's about it.  I don't swear that it's perfect, but if you do find
problems or end up improving it, please share. :)

sfrost@snowman.net